Position evaluation

[Robert Pope]

I had the following position in HGM's tournament last weekend, and KingSlayer and Abbess had pretty different evaluations. What would you score this position at, and why?

[d]r3k2r/p3qpbp/1np1bp2/8/2p1PB2/2N5/RP1NBPPP/2Q1K2R w Kkq - 6 22
White is up a knight and down a pawn, but KingSlayer scored white at over 4 pawns, while Abbess was closer to 2.0. Crafty agrees with the higher score, but I'm not sure why.

[Michael Sherwin]

Blacks horrible pawn structure. Black has poor king safety. Whites extra piece. Whites rook on an open file. Whites centralization of pieces. Blacks pieces uncoordinated which might lead to further material gain in the search. Some top programs have a command to display all the eval terms. Sorry I do not remember which ones.

[Robert Pope]

Okay, I see black's doubled pawns. But is the King safety any worse for black, compared to white? I thought they were equivalent, or maybe even black being a little better.

[JVMerlino]

I had the following position in HGM's tournament last weekend, and KingSlayer and Abbess had pretty different evaluations. What would you score this position at, and why?

[d]r3k2r/p3qpbp/1np1bp2/8/2p1PB2/2N5/RP1NBPPP/2Q1K2R w Kkq - 6 22
White is up a knight and down a pawn, but KingSlayer scored white at over 4 pawns, while Abbess was closer to 2.0. Crafty agrees with the higher score, but I'm not sure why.

This is Crafty's full output:

Code: Select all

                        +-----------white----------+-----------black----------+
material.......   3.18  |    comp     mg      eg   |    comp     mg      eg   |
pawns..........   1.27  |   -0.01   -0.02    0.06  |    1.28    1.27    1.48  |
passed pawns...   0.00  |    0.00    0.00    0.00  |    0.00    0.00    0.00  |
knights........   0.14  |    0.08    0.08    0.08  |    0.06    0.06    0.06  |
bishops........   0.18  |    0.80    0.79    1.15  |   -0.62   -0.61   -0.97  |
rooks..........   0.04  |   -0.02   -0.02   -0.02  |    0.06    0.06    0.06  |
queens.........  -0.07  |   -0.01   -0.01   -0.01  |   -0.06   -0.06   -0.06  |
kings..........   0.14  |   -0.44   -0.45   -0.40  |    0.58    0.59    0.40  |
development....   0.00  |   -0.57   -0.60    0.00  |    0.57    0.60    0.00  |
pawn races.....   0.00  +--------------------------+--------------------------+
total..........   4.86

My (fairly crappy) engine scores this position as +3.6 for White.

[kgburcham]

[D] r3k2r/p3qpbp/1np1bp2/8/2p1PB2/2N5/RP1NBPPP/2Q1K2R w Kkq -


Engine: Houdini 6.01 Pro x64-pext (8192 MB)
by Robert Houdart
8 threads used

11/23 0:00 +4.96 1.Ra5 O-O 2.O-O Rfd8 3.h3 Nd7 4.Nxc4 Nc5 5.Bd6 Rxd6 6.Rxc5 (1.509.356) 13721
11/23 0:00 +5.24 1.Qc2 Rd8 2.O-O Rd7 3.Rfa1 O-O 4.Nf1 Rfd8 5.Ne3 Rd2 6.Qc1 (2.181.499) 13221
12/26 0:00 +5.27 1.Qc2 O-O 2.O-O Nd7 3.Nxc4 Ne5 4.Ra4 Rfd8 5.Ne3 Qb7 6.Nc4 Bxc4 (3.078.357) 12412
13/27 0:00 +5.35 1.Qc2 O-O 2.O-O Rfd8 3.Rfa1 Qb7 4.Be3 Bf8 5.f4 Bh6 6.Rf1 Bf8 7.Ra5 (4.585.027) 12527
14/27 0:00 +5.42 1.Qc2 O-O 2.O-O Rfd8 3.Rfa1 f5 4.exf5 Nd5 5.Nxd5 Bxd5 6.Bxc4 Bd4 7.Qd3 Bxc4 (5.860.956) 12824
15/32 0:00 +5.39 1.Qc2 O-O 2.O-O Nd7 3.Nxc4 Ne5 4.Ra4 Rad8 5.Be3 Bxc4 6.Bxc4 Ng4 7.Bf4 Ne5 8.h3 (8.789.926) 13297
16/35 0:01 +5.41 1.Ra5 O-O 2.O-O Qb4 3.Qa1 Nd7 4.Nf3 Rab8 5.Bxb8 Rxb8 6.Nd4 f5 7.Nxc6 Qxb2 8.exf5 Qxa1 9.Rfxa1 Bxc3 10.fxe6 Bxa1 11.exf7+ (21.126.416) 13691
17/41 0:02 +5.38 1.Qc2 O-O 2.O-O Nd7 3.Nxc4 Ne5 4.Ra4 Qb7 5.Na5 Qc8 6.Nb3 Ng6 7.Be3 f5 8.Nc5 f4 9.Ba6 (32.281.716) 13920
18/41 0:04 +5.50 1.Ra5 O-O 2.O-O Rfe8 3.h3 Rad8 4.Rh5 f5 5.exf5 Bd5 6.Be3 Rd7 7.Rg5 Kh8 8.Nf3 Bxf3 9.Bxf3 Qf6 10.Re1 (59.623.730) 14264
19/41 0:06 +5.36 1.Qc2 O-O 2.O-O Rfd8 3.Be3 Bf8 4.Nf3 Qc7 5.Nd4 c5 6.Nf5 Kh8 7.Raa1 Rd3 8.Bxd3 cxd3 9.Qxd3 Bc4 10.Qc2 (90.517.779) 14756
20/41 0:07 +5.36 1.Qc2 O-O 2.O-O Rfd8 3.Be3 Bf8 4.Nf3 Qc7 5.Nd4 Bb4 6.Qc1 Kh8 7.Rd1 a5 8.Bf4 Qb7 9.h3 a4 10.Bh5 c5 11.Nxe6 (119.106.170) 15044
21/45 0:17 +5.41 1.Qc2 O-O 2.O-O a5 3.Be3 Qc7 4.Rd1 Rfd8 5.Rda1 Bf8 6.Rxa5 Rxa5 7.Rxa5 Bb4 8.Ra1 Nd7 9.Nxc4 Bc5 10.Qc1 Bxc4 11.Bxc4 Qb6 12.Bxc5 (270.390.427) 15571
22/49 0:27 +5.47 1.Ra5 Qb4 2.Rh5 Nd7 3.O-O O-O 4.Nf3 Rfb8 5.Bxb8 Rxb8 6.Nd4 Qxb2 7.Nxe6 Qxc1 8.Rxc1 fxe6 9.Bxc4 Kf7 10.Ra1 Rb7 11.Rxh7 Ne5 12.Ba6 Rc7 13.f4 Ng4 14.Bc4 (431.225.786) 15835 TB:1
23/51 0:34 +5.49 1.Ra5 O-O 2.O-O Rfe8 3.Rh5 a5 4.Be3 Nd7 5.Nxc4 Qb4 6.Rd1 Bxc4 7.Bxc4 Qxc4 8.Rxd7 Qe6 9.Rd1 Rad8 10.Rxd8 Rxd8 11.Bb6 Re8 12.Rxa5 Qb3 13.Bc5 (560.675.450) 16114 TB:2
24/51 1:20 +5.67 1.O-O O-O 2.Ra5 Rfe8 3.h3 f5 4.exf5 Bd5 5.Be3 Qb4 6.Qa1 Nd7 7.Rd1 Rab8 8.Bxc4 Bxc4 9.Ra4 Qxb2 10.Qxb2 Rxb2 11.Nxc4 Rbb8 12.Nd6 Red8 13.Nce4 a5 14.Bg5 (1.312.640.715) 16338 TB:24
25/51 1:32 +5.63 1.O-O O-O 2.Ra5 Rfd8 3.h3 Re8 4.Rh5 Rad8 5.Be3 Rd3 6.Nf3 Rxe3 7.Qxe3 f5 8.Rd1 fxe4 9.Qxe4 h6 10.Nd4 Bd7 11.Qf4 Qb4 12.Nf5 Bxf5 13.Rxf5 Qxb2 14.Bh5 (1.531.582.574) 16609 TB:28
26/57 3:49 +5.57 1.O-O Rd8 2.Ra5 O-O 3.Nf3 Kh8 4.h3 Nd7 5.Rxa7 Qb4 6.Nd4 Qb6 7.Be3 c5 8.Nxe6 fxe6 9.Ra1 f5 10.Bxc4 Ne5 11.Be2 Qb8 12.Ra6 Qc8 13.Na4 c4 14.Nc5 (4.089.325.297) 17781 TB:240
27/61 4:26 +5.62 1.O-O O-O 2.Ra5 Qb4 3.Rh5 Rfd8 4.h3 Nd7 5.Nf3 Rac8 6.Rd1 Kh8 7.Rd4 Ne5 8.Rxd8+ Rxd8 9.Nxe5 fxe5 10.Bxe5 f6 11.Bxf6 Bxf6 12.Qh6 Qe7 13.e5 Qg7 14.Qxf6 (4.769.065.348) 17918 TB:303
28/61 6:24 +5.60 1.O-O O-O 2.Ra5 Rad8 3.Nf3 Kh8 4.h3 Nd7 5.Rxa7 Qc5 6.Be3 Qd6 7.Ra5 Rb8 8.Nd4 Ne5 9.Rd1 Rg8 10.Bf4 Qb4 11.Ra2 Qb7 12.Kh2 Rbd8 13.Bg3 Nd3 14.Qe3 (6.978.097.149) 18171 TB:437
29/63 9:20 +5.72 1.Ra5 O-O 2.O-O Rfe8 3.Nf3 Rad8 4.h3 Qf8 5.Be3 Nd7 6.Nd4 Ne5 7.Rd1 Qb4 8.Qa1 Bd7 9.Rxa7 Bf8 10.f4 Nd3 11.Bxd3 cxd3 12.Rxd3 Qb8 13.Nde2 Bc8 14.Ra8 (10.237.925.693) 18256 TB:1.235
30/63 10:46 +5.75 1.Ra5 (11.646.047.809) 18018 TB:1.309

[MikeB]

I had the following position in HGM's tournament last weekend, and KingSlayer and Abbess had pretty different evaluations. What would you score this position at, and why?

[d]r3k2r/p3qpbp/1np1bp2/8/2p1PB2/2N5/RP1NBPPP/2Q1K2R w Kkq - 6 22
White is up a knight and down a pawn, but KingSlayer scored white at over 4 pawns, while Abbess was closer to 2.0. Crafty agrees with the higher score, but I'm not sure why.

very useful site:

https://hxim.github.io/Stockfish-Evaluation-Guide/

just paste the fen in - you will get something like this



Sf evaluates the material here about 3 pawns up , the imbalance, mobility , pawns, king and placement of pieces are worth almost another two pawns. Note that black has no good side to castle , white's piece placement is not that great , but black's piece placement is worst. But the 4 big differences after material are pawns, imbalance, kings and mobility. Each one is worth about ~ 1/2 pawn or so ( some are slightly less than 1/2 pawn).

[MikeB]

MCb vs Mcb - white can spin up too many threats where there not enough defensive resources for black.

[pgn]
[Event "Mac Pro x5690 3.46 Ghz 18 CPU"]
[Site "Mac-Pro.local"]
[Date "2017.10.30"]
[Round "-"]
[White "SF-McBrain v3.0 TCEC-X 6"]
[Black "SF-McBrain v3.0 TCEC-X 6"]
[Result "1-0"]
[TimeControl "300+3"]
[FEN "r3k2r/p3qpbp/1np1bp2/8/2p1PB2/2N5/RP1NBPPP/2Q1K2R w Kkq - 6 1"]
[SetUp "1"]


1. O-O {+6.00/28 +290} O-O {-5.35/25 +287} 2. Qc2 {+6.06/27 +286} a5
{-5.51/27 +268} 3. Rfa1 {+6.07/21 +287} Rfd8 {-5.74/26 +254} 4. Be3
{+6.14/24 +282} Rdb8 {-5.80/28 +232} 5. Nf3 {+6.20/26 +274} Nd7
{-6.00/30 +222} 6. Nd4 {+6.38/27 +263} Qd6 {-6.23/27 +197} 7. f4
{+6.50/26 +261} Nc5 {-6.38/30 +185} 8. f5 {+6.43/25 +263} Bd7
{-6.59/21 +187} 9. Bxc4 {+6.59/26 +261} Bh6 {-6.51/23 +184} 10. Bxh6
{+7.75/25 +259} Qxd4+ {-6.97/31 +177} 11. Kh1 {+8.04/26 +261} Bxf5
{-8.00/31 +162} 12. exf5 {+8.15/30 +258} Qxc4 {-8.37/22 +164} 13. Qf2
{+8.28/24 +259} Kh8 {-8.36/29 +158} 14. Qg3 {+8.44/28 +256} Rg8
{-8.63/24 +158} 15. Qd6 {+8.45/28 +257} Ne6 {-8.56/29 +154} 16. fxe6
{+8.53/33 +252} Rad8 {-8.78/25 +156} 17. Qe7 {+8.57/26 +253} Qxe6
{-8.90/31 +143} 18. Qxe6 {+8.96/27 +245} fxe6 {-9.06/25 +144} 19. Rxa5
{+9.00/24 +246} Rg4 {-9.06/29 +139} 20. Ra7 {+9.19/26 +240} Re8
{-9.26/32 +129} 21. Rc7 {+10.01/31 +222} Rg6 {-9.31/31 +115} 22. Be3
{+10.26/26 +221} f5 {-10.10/32 +97} 23. Raa7 {+10.48/30 +215} e5
{-10.19/31 +93} 24. Rxh7+ {+10.81/29 +213} Kg8 {-7.80/1 +96} 25. Rhf7
{+11.73/27 +215} Rf8 {-12.83/31 +88} 26. Rfe7 {+49.30/28 +212} Kh8
{-24.27/32 +54} 27. Rxe5 {+54.26/35 +202} Rg7 {-54.56/32 +40} 28. Rxg7
{+59.22/30 +198} Kxg7 {-54.78/38 +12} 29. Bd4 {+60.19/31 +197} Kg8
{-80.70/40 -2}
{User adjudication} 1-0
[/pgn]

[MikeGL]

I had the following position in HGM's tournament last weekend, and KingSlayer and Abbess had pretty different evaluations. What would you score this position at, and why?

[d]r3k2r/p3qpbp/1np1bp2/8/2p1PB2/2N5/RP1NBPPP/2Q1K2R w Kkq - 6 22
White is up a knight and down a pawn, but KingSlayer scored white at over 4 pawns, while Abbess was closer to 2.0. Crafty agrees with the higher score, but I'm not sure why.

General rule, according to my NM friend is to just count doubled-pawns as 1, not two. 1pc up + 1pawn up = 4. I am sure he picked up that assessment rule from a strong GM too. So engine score of around +4 is almost correct I think.

[AdminX]

I had the following position in HGM's tournament last weekend, and KingSlayer and Abbess had pretty different evaluations. What would you score this position at, and why?

[d]r3k2r/p3qpbp/1np1bp2/8/2p1PB2/2N5/RP1NBPPP/2Q1K2R w Kkq - 6 22
White is up a knight and down a pawn, but KingSlayer scored white at over 4 pawns, while Abbess was closer to 2.0. Crafty agrees with the higher score, but I'm not sure why.

very useful site:

https://hxim.github.io/Stockfish-Evaluation-Guide/

just paste the fen in - you will get something like this



Sf evaluates the material here about 3 pawns up , the imbalance, mobility , pawns, king and placement of pieces are worth almost another two pawns. Note that black has no good side to castle , white's piece placement is not that great , but black's piece placement is worst. But the 4 big differences after material are pawns, imbalance, kings and mobility. Each one is worth about ~ 1/2 pawn or so ( some are slightly less than 1/2 pawn).

Thanks Mike, That is very useful.

[Nordlandia]

Hard position for most engines.

[d]rnbqkbnr/8/8/pppppppp/PPPPPPPP/8/8/RNBQKBNR w KQkq - 0 1